Skip to Content

In Continuation with previous blog: Matching & Merging: Equals Vs Token Equals (Part 1)

 

Through this blog, I will try my best that this difference between Equals Vs Token Equals is easily understood by everyone and how scores get affected and come different than what we have set in case of Function = Token Equals.

 

Working with function “Token Equals

At this point of time you know that for executing Matching, you need to define Transformation, Rule and Strategy.

Open MDM Data Manager, Go to Matching Mode:  

 

1. Define a Transformation :

image

 

2. Define a Rule : Here in this Rule “Matching Material Description”, I have included Transformation just created above and set Function = Token Equals and other parameters as shown in below screen-shot.

image

 

3. Define a Strategy now : Here in this Strategy, I have included the Rule “Matching Material Description” as created above and set other parameters as shown in below screen-shot.

image

 

Now coming to Data Manager: Record Mode
Now, I have four records for which there is value for field Material Description as shown below:

image

 

Since we have set Rule Property Function = Token Equals, it will treat these 3 tokens as separate (distinct) token. Lets see how it shows score when we execute Strategy 

image

 

After executing the strategy: In matching Mode, we get the following scores 

image

 

Now you are wondering with Material Description “Lexan IP 300” we have the right Score 100 which we defined during Rule but what about other two records how the score 20 is coming for Material Desciption “Lexan ID 900” and how it is showing score 50 for Material Description “Lexan IP 600” since we have not set these scores during defining Rule.

Logic is pretty simple:

1stly in Function “Token Equals” each Token is considered as separate token (identity) unlike Function “Equal” where these 3 tokens considered as single individual token.

2ndly it gives score basis on the below formula:
Score = Success * Number of Unique Matching Tokens / Total Number of Unique Tokens

Now let’s see score for each of these Material Description:

For Material Description : “Lexan IP 300” with “Lexan IP 300”
Number of Unique Matching Tokens: 3 (Lexan, IP and 300)
Total Number of Unique Tokens: 3 (Lexan, IP and 300)

So Score: 100* 3 / 3 = 100.

For Material Description: “Lexan ID 900” with “Lexan IP 300”
Number of Unique Matching Tokens: 1 (Lexan)
Total Number of Unique Tokens: 5 (Lexan, ID, IP, 300 and 900)

So score: 100* 1 / 5 = 20

For Material Description: “Lexan IP 600” with “Lexan IP 300”
Number of Unique Matching Tokens: 2 (Lexan and IP)
Total Number of Unique Tokens: 4 (Lexan, IP, 600 and 300)

So score: 100* 2 / 4 = 50

My primary objective is to show Score calculation through these blogs. So  I hope now you would not wonder if you find Matching score different than you defined in Rule if you are using Property, Function = Token Equals.

References:

MDM Data Manager Reference Guide

To report this post you need to login first.

4 Comments

You must be Logged on to comment or reply to a post.

  1. Balachandar P
    Hi Mandeep.

    Thanks for the Blog. I have some basic questions.

    >How is count calculated? Is my understanding  correct. Since you are executing Selected vs Selected, Is it checking each field value with other field values. (i.e)Record1 with Record 2,3,4,5 or any other way.

    Thanks
    Bala

    (0) 
    1. Mandeep Singh Saini Post author
      Hi Bala,
      First for all thanks for compliment !!

      Your understanding is correct, It is checking each field value with other field values. I mean Record value Lexan IP 300 with other Record values Lexan IP 600, Lexan ID 900 and Lexan IP 300 OR in other words Matches selected records against themselves.

      As i wrote in Blog my primary objective was to show the [score] column calculation but you are seeking into other information too like Selected Vs Selected etc. So you should post a new thread regarding this where i would answer to your query ;-), Just kidding !!

      See, Other than [SCORE] column, [Count] and [Class] are also significant.

      As per my understanding, it gives the exact count as per your threshold values but Shows the Max Score only among the records.

      I elaborate each of these records again:

      For “Lexan ID 900”, when it compares with (Lexan IP 300), you will get score 20.
      and similarly when “Lexan ID 900” compares value with (Lexan IP 600)and gives score 20.
      Max value among these records comparison is 20, So showing score 20.
      Since score 20 is less than minimum threshold value 50, so therefore count is Zero “0”

      For “Lexan IP 300”, when it compares with (Lexan IP 300), you will get score 100.
      and similarly when “Lexan IP 300” compares with (Lexan IP 600), you will get score 50.
      again when “Lexan IP 300” compares with (Lexan ID 900), you will get score 20.
      Max value among these these records comparison is 100, So showing score 100
      Since score 100 is greater than Max threshold value 80 and 50 is equal to minimum Threshold
      value 50 so showing count “2”.

      Similarly for other “Lexan IP 300”.

      For “Lexan IP 600” when it compares with (Lexan IP 300), you will get score 50.
      and similarly when “Lexan IP 600” compares with (Lexan ID 300), you will get score 50.
      again when “Lexan IP 600”  compares with (Lexan ID 900), you will get score 20.
      Max value among these these records comparison is 50, So showing score 50
      since 50 is equal to minimum Threshold values so showing count “2”.

      So far i hope you got familiar with column Score and Count.

      Now coming to Class, It has values as High(Green), Low(Blue) and None(Red)

      High: Records whose total score is greater than or equal to the High Threshold score are included in the list of potential matches in the Matches tab and placed in the High match level

      Low: Records whose total score is greater than or equal to the Low Threshold score are included in the list of potential matches in the Matches tab and placed in the Low match level.

      Hope you got my point 🙂 !!

      Thanks,
      Mandeep

      (0) 
      1. Balachandar P
        Perfectly explained Mandeep.
        Thanks once again.

        I think I made you to reply in the comments section instead of writing you a new blog 😉

        Cheers.
        Bala

        (0) 
  2. Balachandar P
    Perfectly explained Mandeep.
    Thanks once again.

    I think I made you to reply in the comments section instead of writing you a new blog  😉

    Cheers.
    Bala

    (0) 

Leave a Reply